Abstract:

Contrary to the problem of premature and low searching precision which the particle swarm optimization (PSO) has, this paper improved it from two aspects: the method of fixing inertia weight and the method of improving the algorithm’s searching precision. The inertia weight was determined by a function whose value was decreased nonlinearly and a stochastic value. The stochastic value randomness to jump out the local optima is used. In order to improve the particle’s diversity and the algorithm’s ability of searching global optima, the scale chaos searching was introduced. Also we made a comparison with the standard particle swarm optimization (SPSO) with wavelet to forecast foundation settlement. The experiment indicated that the method had strong global and local searching optima and high forecast precision.

Key words: wavelet analysis; particle swarm optimization; foundation settlement; prediction